EP2190206B1 - Dispositif pour la recherche de mouvement dans le codage dynamique d'images - Google Patents

Dispositif pour la recherche de mouvement dans le codage dynamique d'images Download PDF

Info

Publication number
EP2190206B1
EP2190206B1 EP20080831148 EP08831148A EP2190206B1 EP 2190206 B1 EP2190206 B1 EP 2190206B1 EP 20080831148 EP20080831148 EP 20080831148 EP 08831148 A EP08831148 A EP 08831148A EP 2190206 B1 EP2190206 B1 EP 2190206B1
Authority
EP
European Patent Office
Prior art keywords
unit
data
screen
read
memory
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
EP20080831148
Other languages
German (de)
English (en)
Japanese (ja)
Other versions
EP2190206A1 (fr
EP2190206A4 (fr
Inventor
Koyo Nitta
Hiroe Iwasaki
Jirou Naganuma
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nippon Telegraph and Telephone Corp
Original Assignee
Nippon Telegraph and Telephone Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nippon Telegraph and Telephone Corp filed Critical Nippon Telegraph and Telephone Corp
Publication of EP2190206A1 publication Critical patent/EP2190206A1/fr
Publication of EP2190206A4 publication Critical patent/EP2190206A4/fr
Application granted granted Critical
Publication of EP2190206B1 publication Critical patent/EP2190206B1/fr
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
    • H04N19/43—Hardware specially adapted for motion estimation or compensation
    • H04N19/433—Hardware specially adapted for motion estimation or compensation characterised by techniques for memory access
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
    • H04N19/436—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation using parallelised computational arrangements
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
    • H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding

Definitions

  • the present invention relates to a motion search apparatus in video coding for performing motion search by using a systolic array in video coding.
  • the systolic array is a computing apparatus in which a plurality of processor elements (to be referred to as PE hereinafter) are arranged regularly, and calculation target data flows through the PEs like a pipeline so that computational processing is executed by the PEs in parallel and with high speed.
  • the computing apparatus is also called a PE array.
  • the PE array is used in processing for repeating calculation of the sum of absolute difference (SAD) of pixel values between a coding target block of the original image and a reference image within a motion search range of the reference image, so that speed-up of motion vector detection is realized.
  • SAD sum of absolute difference
  • Fig. 9A shows a configuration example of the PE array in a conventional motion search apparatus. This example is configured to calculate the sum of absolute difference between the original image of 4 ⁇ 2 pixels and the reference image by using 8 processor elements PE00 - PE31. By increasing the number of PEs, the PE array can be also configured to calculate the sum of absolute difference in units of 4 ⁇ 4 pixels or 8 ⁇ 8 pixels, for example. In addition, by combining a plurality of PE arrays 40 shown in Fig. 9A , a calculation circuit can be configured for calculating the sum of absolute difference of n ⁇ m pixels (n ⁇ 4, m ⁇ 4).
  • the PE array 40 includes input terminals of original image input data SMB, 4 pieces of reference image input data RA00, RA01, RA10 and RA11, and a reference image switch control input signal RASW for controlling a selector for selecting reference image input data to be calculated.
  • the PE array 40 includes an output terminal for an output ADOUT of a result of accumulation of the sum of absolute difference.
  • each PE includes an input terminal MBin for inputting original image input data, an input terminal ADDin for inputting a sum value from a left adjacent PE, an output terminal ADDout for outputting a sum value to a right adjacent PE, and input terminals RAin0 and RAin1 for inputting two pieces of reference image input data.
  • Figs. 10A - 10C are diagrams for explaining operation of the PE array 40 shown in Fig. 9A .
  • the PE array 40 performs calculation for searching a reference image (pixel values x00, x01, ...) shown in Fig. 10B for a part by which the sum of absolute value becomes the smallest with respect to a 4 ⁇ 2 pixel group (pixel values c00 - c31) of the original image shown in Fig. 10A .
  • pixel values c00, c10 and c20 .... of the original image are sequentially input to PE00, PE10 and PE20 ... in initial 8 cycles (clock CLKs), and held.
  • PE10 calculates a value S10 by adding the absolute difference between a pixel value c10 of the original image and a pixel value x10 of the reference image to the value S00 calculated by the PE00 in cycle 1.
  • PE10 calculates a value S11 by adding the absolute difference between a pixel value c10 of the original image and a pixel value x20 of the reference image to the value S01 calculated by the PE00 in cycle 2.
  • PE20 calculates a value S20 by adding the absolute difference between a pixel value c20 of the original image and a pixel value x20 of the reference image to the value S10 calculated by the PE10 in cycle 2.
  • each of PE00 - PE31 executes calculation like a pipeline, so that a sum of absolute difference between c00-c31 and x00-x31 is output from the output terminal ADOUT of the PE array 40 initially.
  • the sum of absolute difference between c00-c31 and x10-x41 is output, and in the next cycle, the sum of absolute difference between c00-c31 and x20-x51 is output.
  • the sum of absolute difference within the search range of the motion vector is sequentially output in each cycle (refer to non-patent documents 1 and 2, and patent documents 1 and 2 for more details).
  • Fig. 11 shows a timing chart in the PE array 40.
  • HOLDMB indicates a start signal that instructs each of PE00 - PE31 to hold original image input data SMB and to start calculation.
  • CLK indicates a clock
  • HOLDSEL indicates a reference image switch control input signal.
  • a pixel value of the reference image is represented as pixel coordinates (x, y) of the reference image. For example, (0, 0) corresponds to a pixel value x00 shown in Fig. 10 .
  • Each pixel value of the reference image is sequentially supplied to the PE array 40.
  • reference image pixel values of (0, 0) - (6, 0) and (0, 1) - (6, 1) are simultaneously input from the reference image memory at CLK0
  • reference image pixel values of (0, 2) - (6, 2) and (0, 3) - (6, 3) are simultaneously input from the reference image memory at CLK8.
  • first 7 pixels are sequentially supplied to the PE array 40 in 7 clocks starting from CLK1, and remaining 7 pixels are sequentially supplied to the PE array 40 in 7 clocks starting from CLK5.
  • CLK9 a result of accumulation of the sum of absolute difference at the search origin position coordinates (0, 0) is output.
  • Fig. 12 shows read timing for reading data from the reference image memory.
  • the pieces of data of 7 pixels x 2 are read from the reference image memory-Since two pieces of data cannot be simultaneously read from a same memory bank, data is stored in each of banks like Bank0, Bank1, Bank2, Bank0, ..., for each line of the image, for example. Accordingly, bank conflict can be avoided, so that pieces of data in Bank0 and Bank1 can be read simultaneously from the reference image memory, for example.
  • Figs. 13A - 13C are diagrams showing a search range for detecting a motion vector.
  • the motion search is performed for searching the reference image 100 for a position having a pixel value group that is the closest to the coding target block 201 in the original image 200 shown in Fig. 13A .
  • a pixel value group of the size of the coding target block is sequentially cut out from the search region 102 of a predetermined size while shifting little by little centering around a search center 101 that is the coordinate position of the coding target block, so that the sum of absolute difference (SAD) between the cut out pixel values and the pixel values of the coding target block 201 is calculated.
  • SAD sum of absolute difference
  • the H.264 coding scheme is defined such that the motion search can be performed by extending to the outside of the reference image 100. That is, in the specification of the H.264, as shown in Fig. 13B , when the position of the coding target block 201 is near an end part of the original image 200, the search region 100 extends to the outside of the reference image 100.
  • Fig. 13C as to pixel values of the outside of the screen of the reference image 100, values copied from pixel values of the uppermost stage of the reference image 100 is used for the outside of the upper side of the screen, and values copied from pixel values of the leftmost line of the reference image 100 is used for the outside of the left side of the screen. Copied values used for the lowermost stage and the rightmost end of the reference image 100 are used similarly.
  • the size of the reference image memory for one reference image is set to be larger than the size of the reference image by the amount of the search region of the outside of the screen, and the pixel values of the outside of the screen is set to be the same as the pixel values of the end part of the screen beforehand when storing the reference image.
  • this method has a problem in that additional hardware amount for the reference image memory becomes necessary.
  • Figs. 14A and 14B are diagrams for explaining a memory bank configuration and explaining the problem of bank conflict when reading pixels outside the screen.
  • LAX indicates logical address in the X direction
  • LAY indicates logical address in the Y direction.
  • #0, #1, ... , #5 indicate bank numbers of memory banks. When bank numbers are different between memory banks, data can be simultaneously read from the memory banks. But, data in areas of the same bank number cannot be read simultaneously due to bank conflict.
  • An object of the present invention is to solve the above-mentioned problem, and to provide a motion search apparatus in which bank conflict does not occur without increasing memory for storing the pixels outside the screen.
  • the uppermost line in the screen is read. Further, only the lowermost line is read in advance simultaneously with reading of the next previous line so that the read timing is advanced, in order to avoid bank conflict when reading the lowermost line in a search range and the uppermost line in a next search range simultaneously.
  • the data of the lowermost line read in advance is held in a data holding circuit such as a shift register and the like, and is supplied from the data holding circuit when the PE array needs the data for calculation.
  • pixels of the search range are outside the screen of the reference image in a horizontal direction
  • pixel values of the end of the screen are copied after reading the line.
  • the line of the end of the screen that has been read is copied and can be used as it is.
  • the present invention is a motion search apparatus for use in video coding, including:
  • the pixel permuting unit includes a vertical direction copy unit configured, when a region of the motion search range is a region outside the screen of the reference image in a vertical direction, to copy pixel data of a line of a boundary part of the screen of the reference image as pixel data of the region outside the screen, and to input the copied data to the PE array unit.
  • the pixel permuting unit includes a horizontal direction copy unit configured, when a region of the motion search range is a region outside the screen of the reference image in a horizontal direction, to copy pixel data of a boundary part of the screen in the same line of the reference image in the horizontal direction as pixel data of the region outside the screen, and to input the copied data to the PE array unit.
  • a motion vector outside the screen can be also detected. Especially, it is not necessary to have a memory for the outside (vertical direction) of the screen, and also, it is not necessary to have a memory for the outside (horizontal direction) of the screen. In addition, bank conflict can be avoided without increasing the bank number of memory banks.
  • Figs. 1A and 1B are diagrams for explaining the outline of an embodiment of the present invention.
  • data 3 of the lowermost line in the search range is stored in a bank a of the memory, and that data 1 and data 2 of the uppermost line of a next search range are stored in bank a and bank b respectively. If these pieces of data are read simultaneously from the reference image memory 30 in order to supply them to the PE array unit 4, bank conflict occurs since data 1 and data 3 exist in the same bank a.
  • a conflict bank anticipatory read control unit 10 in a memory read unit 1 shown in Fig. 1B performs control to read the data 3 of the lowermost line in advance simultaneously with reading a previous line if bank conflict between the lowermost line and two uppermost lines in the search range occurs.
  • a pixel permuting unit 2 permutes pixel data read from the memory 3 according to calculation order in the PE array unit 4 and provides the permuted data to the PE array unit 4. Especially, the pixel permuting unit 2 holds the data 3 of the lowermost line read in advance in a read data holding circuit 20, and provides the data to the PE array unit 4 at a time when it becomes necessary to input the data to the PE array unit 4.
  • a horizontal direction copy unit 22 copies pixel values of the end of the screen and provides the copied data to the PE array unit 4.
  • a vertical direction copy unit 24 copies pixel values of a boundary of the upper side of the screen or a boundary of the lower side of the screen and provides the copied data to the PE array unit 4.
  • Fig. 2 is a whole block diagram of the motion search apparatus of the present embodiment.
  • Fig. 3 is a detailed block diagram of the memory read unit 1
  • Fig. 4 is a detailed block diagram of the pixel permuting unit 2.
  • Configuration of each unit other than the memory read unit 1 and the pixel permuting unit 2 is similar to configuration of the conventional motion search apparatus using systolic array.
  • the memory 3 stores pixel data of the original image and the reference image.
  • the pixel data of the original image may be stored in another memory and may be input from an input terminal.
  • the PE array unit 4 is configured by a plurality of PE arrays 40 according to the size of the coding target block, each PE array 40 including a plurality of processor elements (PE), selector, adding circuit, and register and the like as described by referring to Figs. 9A and 9B .
  • An evaluation unit 5 evaluates the motion vector based on the sum of absolute difference (SAD) output from the PE array unit 4, and cost information of coding such as difference between motion vector MV and predicted motion vector PMV, and returns an evaluation result to the control unit 6.
  • SAD sum of absolute difference
  • the control unit 6 performs control of the whole of the motion search apparatus using program control.
  • the control unit 6 instructs the memory read unit 1 to start search by specifying a search region (center motion vector MV).
  • the memory read unit 1 converts each logical address indicating the search range into each real address in the memory 3, so as to request memory read for the memory 3.
  • the control unit 6 checks whether the search range is outside the screen, and when the range is outside the screen, the control unit 6 reports outside-screen information to the pixel permuting unit 2.
  • the pixel permuting unit 2 performs cut-out and permutation for the pixel data read from the memory 3, and provides the processed data to the PE array unit 4.
  • the memory read unit 1 includes a logical address generation unit 11, a outside-screen determination unit 12, a logical address/real address conversion unit 13, and a memory access unit 14.
  • the logical address generation unit and the logical address/real address conversion unit 13 are similar to those of the conventional apparatus.
  • the logical address generation unit 11 generates a logical address LAX and LAY indicating a relative address of the search range shown in Figs. 14A and 14B based on information of the center motion vector MV that specifies a search region specified by the control unit 6.
  • the outside-screen determination unit 12 determines whether the logical address generated by the logical address generation unit 11 is within the screen of the reference image or outside the screen, and when it is outside the screen, the outside-screen determination unit 12 sends outside-screen information to the memory access unit 14 and to the pixel permuting unit 2.
  • the logical address/real address conversion unit 13 converts the logical address generated by the logical address generation unit 11 to the real address where actual pixel data is stored in the memory 3. The conversion from the logical address to the real address can be easily performed by storing the head real address of the reference image when storing the reference image into the memory 3. Also as to the address of the coding target block in the original image reported from the control unit 6 beforehand, conversion into the real address is performed.
  • the converted real address is sent to the memory access unit 14.
  • the memory access unit 14 sends a read request to the memory 3 at an after-mentioned timing according to Fig. 8 .
  • the conflict bank anticipatory read control unit 10 performs access control to avoid bank conflict as explained in Fig. 1 .
  • the pixel permuting unit 2 includes a horizontal direction (X direction) cutting unit 21, a horizontal direction copy unit 22, a parallel/serial conversion unit 23, a vertical direction (Y direction) copy unit 24, and a vertical direction permuting unit 25.
  • the horizontal direction (X direction) cutting unit 21, the parallel/serial conversion unit 23 and the vertical direction permuting unit 25 are similar to those of the conventional apparatus.
  • the data read from the memory 3 to the pixel permuting unit 2 includes data of 16 pixels per 1 line.
  • the horizontal direction cutting unit 21 cuts out data of 8 pixels ⁇ 3 lines that are necessary for motion search from among the read data.
  • the horizontal direction copy unit 22 When performing calculation for detecting outside screen motion vector in the horizontal direction, the horizontal direction copy unit 22 performs processing for copying pixel values of the boundary of the screen (screen end) as pixel data outside the screen based on the outside-screen information reported from the outside-screen determination unit 12.
  • Figs. 5A and 5B are diagrams for explaining a method for copying the pixel values performed by the horizontal direction copy unit 22.
  • EILRFLG indicates a left and right region flag. When the flag is "00", the whole search region exists within the screen. When the flag is "10”, the end of the screen exists in the left side, and when the flag is "01", the end of the screen exists in the right side.
  • EIXLVAL indicates a distance from the left side screen end in the X direction
  • EIXRVAL indicates a distance from the right side screen end in the X direction.
  • D0 - D7 are data of pixel values
  • MSB indicates data at the uppermost side of an 8 pixel data sequence
  • LSB indicates the lowermost data.
  • Fig. 6 shows a circuit configuration example of the horizontal direction copy unit 22.
  • EDI0 - EDI7 are input data of pixel values of D0 D7 respectively shown in Figs. 5A and 5B .
  • XEIXCPO0 - XEIXCPO7 are X direction pixel copy process data 0 - X direction pixel copy process data 7 respectively, and are output data of the result obtained by copying the screen end data into the screen outside as shown in Fig. 5 .
  • the horizontal direction copy unit 22 selects input data EDI0 - EDI7 of pixel values using selectors shown in Fig. 6 based on input signals of EILRFLG, EIXLVAL and EIXRVAL, so that the horizontal direction copy unit 22 performs copy of data in the X direction as shown in Figs. 5A and 5B .
  • the parallel/serial conversion unit 23 shown in Fig. 4 cuts out pixel data of 8 pixel x 3 lines output by the horizontal direction copy unit 22 pixel by pixel for each line in order, converts the cut-out data into pixel data of 1 pixel ⁇ 3 lines, and output the data.
  • the vertical direction copy unit 24 performs processing for copying pixel values of the boundary of the upper side of the screen or the boundary of the lower side of the screen.
  • Fig. 7 shows a circuit configuration example of the vertical direction copy unit 24.
  • EDI0 - EDI2 are input data 0 - 3 of 3 lines.
  • EIUDFLG is an upper/lower region flag indicating whether the region is in the upper side of the screen or in the lower side of the screen, and is a signal received from the outside-screen determination unit 12 as the outside-screen information. When the flag is "00", it indicates that the region is within the screen.
  • EIYSFTSEL is an input data Y direction shift control signal, and indicates whether to shift the input data in vertical direction (Y direction) for copying.
  • EITRG is a memory enable signal, and indicates a timing when reading data from the memory 3.
  • XEIYCPOO - XEIYCPO2 are Y direction pixel copy process data outputs 0 - 2 of 3 lines. In a normal search, two lines of data among them are used. But, when the process goes from the lowermost line of a search range to the uppermost line of a next search range, data of 3 lines are output.
  • the timing generation circuit 241 is a circuit for controlling data holding into the read data holding circuit 20. As explained with reference to Figs. 1A and 1B , the read data holding circuit 20 holds data of the lowermost line read in advance for 8 cycles, and delays the data by 8 cycles and outputs the data.
  • the read data holding circuit 20 is configured by a shift register of 8 stages, for example, and delays the data of the lowermost line of the search range read beforehand by 8 cycles based on the shift enable signal of the timing generation circuit 241.
  • the vertical direction permuting unit 25 permutes the pixel data of 1 pixel ⁇ 3 lines output by the vertical direction copy unit 24 and outputs the data to the PE array unit 4.
  • the processing of the permutation of the pixel data by the vertical direction permuting unit 25 is similar to permutation processing in the conventional motion search apparatus used for MPEG-2 and the like.
  • Fig. 8 shows timing for reading reference image data from the memory 3 in the present embodiment.
  • Reading of data until cycle 24 (CLK 24) and supply timing of data to the PE array are similar to timing in the conventional motion search apparatus described with reference to Fig. 12 .
  • the conventional apparatus shown in Fig. 12 reads pixel data of two lines that are data of bank #0 from (0, 6) to (6, 6) and data of bank #1 from (0, 7) to (6, 7) simultaneously.
  • the apparatus reads pixel data of total 3 lines simultaneously that are data in bank #2 of the lowermost line from (0, 8) to (6, 8), data of banks #0 and #3 of the uppermost line from (4, 0) to (10, 0), and data of banks #1 and #4 of the next uppermost line from (4, 1) to (10, 1).
  • Such a process can be performed since bank conflict does not occur in reading after the 32nd cycle.
  • reading is performed as follows.
  • the apparatus of the present embodiment reads pixel data of two lines that are data of bank #0 from (0, 6) to (6, 6) and data of bank #1 from (0, 7) to (6, 7), and at the same time, the apparatus reads pixel data of the lowermost line of bank #2 from (0, 8) to (6, 8).
  • the pixel data read from the lowermost line is held in the read data holding circuit 20 in order to output to the PE array after 8 cycles.
  • the apparatus reads pixel data of two lines simultaneously that are data of the uppermost line of banks #0 and #3 from (4, 0) to (10, 0), and data of the next uppermost line of banks #1 and #4 from (4, 1) to (10, 1) from the memory 3, and the apparatus outputs, to the PE array, pixel data from (4, 0) to (10, 0) and pixel data from (0, 8) to (6, 8) in units of pixels for each cycle.
  • the apparatus further outputs pixel data of the line from (4, 1) to (10, 1) to the PE array.
  • the above-mentioned motion search apparatus can be configured by an LSI.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Computing Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

La présente invention concerne une technique permettant de réaliser une opération à grande vitesse dans une recherche de mouvement au moyen d'un réseau PE en évitant un conflit entre bancs sans accroître une mémoire utilisée pour stocker des éléments hors-écran. Lorsque des données de pixels dans une pluralité de lignes lues à partir d'une mémoire ((3) mémoire d'images de référence (30)) sont dans le même banc, une unité de contrôle de pré-lecture de banc conflictuel (10) effectue une lecture préalable des données de pixels provenant d'une des lignes et un circuit de maintien de données lues (20) maintient les données jusqu'à une synchronisation d'entrée vers une unité de réseau PE (4). Ainsi, il est possible d'éviter un conflit entre bancs lors d'une lecture de données de pixels provenant de la mémoire (3), réalisant ainsi un traitement pipeline uniforme par l'unité de réseau PE (4).

Claims (4)

  1. Appareil de recherche de mouvement destiné à être utilisé dans le codage vidéo, comprenant:
    une unité de réseau de PE configurée pour calculer une somme de différence absolue entre une image d'origine et une image de référence en effectuant un calcul parallèlement à l'utili-sation d'une pluralité d'éléments processeurs (PE) ;
    une mémoire (30) configurée pour stocker des données de pixels de l'image de référence qui est une cible pour la recherche de mouvement ;
    une unité de lecture de mémoire (1) configurée pour lire les données de pixels de l'image de référence depuis la mémoire ;
    une unité de permutation de pixels (2) configurée pour permuter les données de pixels lues depuis la mémoire et pour entrer les données de pixels dans l'unité de réseau de PE ;
    une unité d'évaluation (5) configurée pour détecter et pour évaluer un vecteur de mouvement dans une plage de recherche prédéfinie de l'image de référence sur la base de la somme de différence absolue calculée par l'unité de réseau de PE ; et
    une unité de commande (6) configurée pour commander la recherche de mouvement par chacune des unités,
    caractérisé en ce que l'unité de lecture de mémoire comprend une unité de commande de lecture par anticipation de banc conflictuel (10) configurée pour effectuer une commande afin, lorsque des éléments de données de pixels d'une pluralité de lignes à entrer simultanément dans l'unité de réseau de PE sont stockées dans un même banc de la mémoire, de déplacer le moment de lecture destiné à lire les données de pixels de la pluralité de lignes depuis la mémoire en unités de lignes, et de lire les données de pixels à un moment précédant un moment requis pour l'entrée dans l'unité de réseau de PE, et
    l'unité de permutation de pixels comprend un circuit de maintien de données lues (20) configuré pour maintenir les données de pixels d'une ligne lue par l'unité de lecture de mémoire au moment précédant le moment requis pour l'entrée dans l'unité de réseau de PE jusqu'au moment de l'entrée dans l'unité de réseau de PE.
  2. Appareil de recherche de mouvement destiné à être utilisé dans le codage vidéo selon la revendication 1, l'unité de permutation de pixels comprenant :
    une unité de copie dans la direction verticale (24) configurée, lorsqu'une région de la plage de recherche de mouvement est une région située à l'extérieur de l'écran de l'image de référence dans une direction verticale, pour copier les données de pixels d'une ligne d'une partie-limite de l'écran de l'image de référence en tant que données de pixels de la région située à l'extérieur de l'écran, et pour entrer les données copiées dans l'unité de réseau de PE.
  3. Appareil de recherche de mouvement destiné à être utilisé dans le codage vidéo selon la revendication 1, l'unité de permutation de pixels comprenant :
    une unité de copie dans la direction horizontale (22) configurée, lorsqu'une région de la plage de recherche de mouvement est une région située à l'extérieur de l'écran de l'image de référence dans une direction horizontale, pour copier les données de pixels d'une partie-limite de l'écran sur la même ligne de l'image de référence dans la direction horizontale en tant que données de pixels de la région située à l'extérieur de l'écran, et pour entrer les données copiées dans l'unité de réseau de PE.
  4. Appareil de recherche de mouvement destiné à être utilisé dans le codage vidéo selon la revendication 2, l'unité de permutation de pixels comprenant :
    une unité de copie dans la direction horizontale configurée, lorsqu'une région de la plage de recherche de mouvement est une région située à l'extérieur de l'écran de l'image de référence dans une direction horizontale, pour copier les données de pixel d'une partie-limite de l'écran sur la même ligne que l'image de référence dans la direction horizontale en tant que données de pixels de la région située à l'extérieur de l'écran, et pour entrer les données copiées dans l'unité de réseau de PE.
EP20080831148 2007-09-13 2008-09-12 Dispositif pour la recherche de mouvement dans le codage dynamique d'images Active EP2190206B1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2007237534A JP4695124B2 (ja) 2007-09-13 2007-09-13 動画像符号化における動き探索装置
PCT/JP2008/066557 WO2009035088A1 (fr) 2007-09-13 2008-09-12 Dispositif pour la recherche de mouvement dans le codage dynamique d'images

Publications (3)

Publication Number Publication Date
EP2190206A1 EP2190206A1 (fr) 2010-05-26
EP2190206A4 EP2190206A4 (fr) 2010-12-29
EP2190206B1 true EP2190206B1 (fr) 2012-05-16

Family

ID=40452094

Family Applications (1)

Application Number Title Priority Date Filing Date
EP20080831148 Active EP2190206B1 (fr) 2007-09-13 2008-09-12 Dispositif pour la recherche de mouvement dans le codage dynamique d'images

Country Status (6)

Country Link
US (1) US8428137B2 (fr)
EP (1) EP2190206B1 (fr)
JP (1) JP4695124B2 (fr)
KR (1) KR101091054B1 (fr)
CN (1) CN101796845B (fr)
WO (1) WO2009035088A1 (fr)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP5360688B2 (ja) * 2009-07-21 2013-12-04 日本電気株式会社 動画復号システム、動画復号装置、参照ブロック生成出方法、及び参照ブロック生成プログラム
TWI523519B (zh) * 2011-01-28 2016-02-21 佳能企業股份有限公司 影像補償方法及系統
JP2019091191A (ja) 2017-11-14 2019-06-13 ルネサスエレクトロニクス株式会社 半導体装置、データ処理システム、データ読取り方法、及びデータ読取りプログラム
US10679320B1 (en) * 2018-07-23 2020-06-09 Ambarella International Lp High dynamic range sensor system with row increment operation
US20230199199A1 (en) * 2021-12-16 2023-06-22 Mediatek Inc. Video Encoding Parallelization With Time-Interleaving Cache Access

Family Cites Families (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0728716B2 (ja) 1989-10-09 1995-04-05 倉敷紡績株式会社 攪拌装置
US5229866A (en) * 1990-10-10 1993-07-20 Fuji Xerox Co., Ltd. Image processing system for reducing wait time before image data transfer to an image output section
US5598514A (en) * 1993-08-09 1997-01-28 C-Cube Microsystems Structure and method for a multistandard video encoder/decoder
FR2719398B1 (fr) * 1994-04-27 1996-07-19 Sgs Thomson Microelectronics Dispositif et procédé d'adressage d'une mémoire cache d'un circuit de compression d'images mobiles.
KR100440522B1 (ko) * 1996-08-29 2004-10-15 마츠시타 덴끼 산교 가부시키가이샤 화상메모리특유의하드웨어스펙에기인하는각종지연요인을저장방식및판독방식의개선에의해극복한화상복호장치및화상메모리
JP3787823B2 (ja) * 1997-07-31 2006-06-21 ソニー株式会社 画像処理装置および画像処理方法
JP3569143B2 (ja) * 1997-12-25 2004-09-22 松下電器産業株式会社 画像データ圧縮伸長処理装置
US6339656B1 (en) * 1997-12-25 2002-01-15 Matsushita Electric Industrial Co., Ltd. Moving picture encoding decoding processing apparatus
US6163576A (en) * 1998-04-13 2000-12-19 Lsi Logic Corporation Video encoder having reduced memory bandwidth requirements
JP3865541B2 (ja) * 1999-08-23 2007-01-10 沖電気工業株式会社 画像照合装置
JP2004080583A (ja) * 2002-08-21 2004-03-11 Renesas Technology Corp 画像処理装置
JP4332013B2 (ja) * 2003-10-28 2009-09-16 日本電信電話株式会社 擬似多ポートメモリ装置および擬似多ポートメモリ装置の画素読み出し方法
JP4419608B2 (ja) * 2004-02-27 2010-02-24 セイコーエプソン株式会社 動画像符号化装置
KR100621137B1 (ko) * 2004-02-27 2006-09-13 세이코 엡슨 가부시키가이샤 동화상 부호화 장치 및 동화상 처리장치
WO2005104027A2 (fr) 2004-04-22 2005-11-03 Koninklijke Philips Electronics N.V. Dispositif de traitement de donnees offrant un acces parallele a des reseaux multidimensionnels de valeurs de donnees
JP4594688B2 (ja) * 2004-06-29 2010-12-08 オリンパス株式会社 画像符号化処理方法、画像復号化処理方法、動画圧縮処理方法、動画伸張処理方法、画像符号化処理プログラム、画像符号化装置、画像復号化装置、画像符号化／復号化システム、拡張画像圧縮伸張処理システム
KR100668302B1 (ko) * 2004-07-28 2007-01-12 삼성전자주식회사 비디오 디코더/엔코더에서의 메모리 맵핑 장치 및 그 방법
JP4690424B2 (ja) * 2005-12-26 2011-06-01 パナソニック株式会社 コマンド処理装置、方法、及び集積回路装置

Also Published As

Publication number Publication date
JP4695124B2 (ja) 2011-06-08
EP2190206A1 (fr) 2010-05-26
KR101091054B1 (ko) 2011-12-08
US20100215105A1 (en) 2010-08-26
CN101796845B (zh) 2012-05-23
US8428137B2 (en) 2013-04-23
KR20100049620A (ko) 2010-05-12
JP2009071569A (ja) 2009-04-02
WO2009035088A1 (fr) 2009-03-19
EP2190206A4 (fr) 2010-12-29
CN101796845A (zh) 2010-08-04

Similar Documents

Publication Publication Date Title
US20090110077A1 (en) Image coding device, image coding method, and image coding integrated circuit
US7835441B2 (en) Supporting motion vectors outside picture boundaries in motion estimation process
JP5059058B2 (ja) 高速動き探索装置及びその方法
KR100267476B1 (ko) 매크로 블록판독 및 설정방법
EP2190206B1 (fr) Dispositif pour la recherche de mouvement dans le codage dynamique d'images
US20050190609A1 (en) Memory interface and data processing system
KR101520027B1 (ko) 움직임 추정 방법 및 장치
KR101617492B1 (ko) 픽셀 레이트에서의 이미지 처리를 위한 방법 및 장치
KR0160618B1 (ko) 실시간 움직임 추정장치 및 그 방법
US6850569B2 (en) Effective motion estimation for hierarchical search
JP4597103B2 (ja) 動きベクトル探索方法及び装置
JP4377693B2 (ja) 画像データ検索
US6411652B1 (en) Motion estimation
US20100074336A1 (en) Fractional motion estimation engine
JP3832431B2 (ja) 画像処理装置
JP3041658B2 (ja) 高並列動き補償演算器
JP2004229150A (ja) 動きベクトル探索方法および装置
US20070109875A1 (en) Data storage method and information processing device using the same
JP3698501B2 (ja) 動きベクトル検出装置
JP4170173B2 (ja) ブロックマッチング演算装置
JP3374014B2 (ja) 画像相関器
JP4332013B2 (ja) 擬似多ポートメモリ装置および擬似多ポートメモリ装置の画素読み出し方法
JPH09503324A (ja) 画素ブロックの比較のためのプロセッサ（ブロックマッチングプロセッサ）
JP3988522B2 (ja) データ処理装置及び半導体装置
US20130094586A1 (en) Direct Memory Access With On-The-Fly Generation of Frame Information For Unrestricted Motion Vectors

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

17P Request for examination filed

Effective date: 20100226

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MT NL NO PL PT RO SE SI SK TR

AX Request for extension of the european patent

Extension state: AL BA MK RS

DAX Request for extension of the european patent (deleted)
A4 Supplementary search report drawn up and despatched

Effective date: 20101125

RIC1 Information provided on ipc code assigned before grant

Ipc: H04N 7/26 20060101ALI20101119BHEP

Ipc: H04N 7/32 20060101AFI20090401BHEP

REG Reference to a national code

Ref country code: DE

Ref legal event code: R079

Ref document number: 602008015724

Country of ref document: DE

Free format text: PREVIOUS MAIN CLASS: H04N0007320000

Ipc: H04N0007500000

GRAP Despatch of communication of intention to grant a patent

Free format text: ORIGINAL CODE: EPIDOSNIGR1

RIC1 Information provided on ipc code assigned before grant

Ipc: H04N 7/26 20060101ALI20111201BHEP

Ipc: H04N 7/50 20060101AFI20111201BHEP

GRAS Grant fee paid

Free format text: ORIGINAL CODE: EPIDOSNIGR3

GRAA (expected) grant

Free format text: ORIGINAL CODE: 0009210

AK Designated contracting states

Kind code of ref document: B1

Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MT NL NO PL PT RO SE SI SK TR

REG Reference to a national code

Ref country code: GB

Ref legal event code: FG4D

REG Reference to a national code

Ref country code: CH

Ref legal event code: EP

REG Reference to a national code

Ref country code: AT

Ref legal event code: REF

Ref document number: 558549

Country of ref document: AT

Kind code of ref document: T

Effective date: 20120615

REG Reference to a national code

Ref country code: IE

Ref legal event code: FG4D

REG Reference to a national code

Ref country code: DE

Ref legal event code: R096

Ref document number: 602008015724

Country of ref document: DE

Effective date: 20120719

REG Reference to a national code

Ref country code: NL

Ref legal event code: VDEP

Effective date: 20120516

REG Reference to a national code

Ref country code: LT

Ref legal event code: MG4D

Effective date: 20120502

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: SE

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: LT

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: CY

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: FI

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: PL

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: IS

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120916

Ref country code: NO

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120816

REG Reference to a national code

Ref country code: AT

Ref legal event code: MK05

Ref document number: 558549

Country of ref document: AT

Kind code of ref document: T

Effective date: 20120516

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: HR

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: LV

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: PT

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120917

Ref country code: SI

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: GR

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120817

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: BE

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: RO

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: NL

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: AT

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: CZ

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: DK

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: SK

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

Ref country code: EE

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: IT

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

PLBE No opposition filed within time limit

Free format text: ORIGINAL CODE: 0009261

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT

26N No opposition filed

Effective date: 20130219

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: ES

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120827

Ref country code: MC

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20120930

REG Reference to a national code

Ref country code: CH

Ref legal event code: PL

REG Reference to a national code

Ref country code: DE

Ref legal event code: R097

Ref document number: 602008015724

Country of ref document: DE

Effective date: 20130219

REG Reference to a national code

Ref country code: IE

Ref legal event code: MM4A

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: CH

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20120930

Ref country code: BG

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120816

Ref country code: LI

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20120930

Ref country code: IE

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20120912

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: MT

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: TR

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20120516

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: LU

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20120912

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: HU

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20080912

REG Reference to a national code

Ref country code: FR

Ref legal event code: PLFP

Year of fee payment: 9

REG Reference to a national code

Ref country code: FR

Ref legal event code: PLFP

Year of fee payment: 10

REG Reference to a national code

Ref country code: FR

Ref legal event code: PLFP

Year of fee payment: 11

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: DE

Payment date: 20240918

Year of fee payment: 17

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: GB

Payment date: 20240919

Year of fee payment: 17

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: FR

Payment date: 20240925

Year of fee payment: 17